Built a production-ready AI cybersecurity platform that detects malware, analyzes threats in real-time, and prevents phishing attacks using machine learning and automated reporting.
The system accepts file and URL submissions through a Next.js web interface, processes them instantly using XGBoost classification models, and generates downloadable threat analysis reports for every detected malicious file.
Real-time threat monitoring runs continuously in the background, flagging suspicious activity the moment it is detected — no manual scanning required.
Built for scale and production deployment with Firebase backend handling data storage and Node.js managing server-side operations.
Tech Stack: Python · Flask · XGBoost · Firebase · Node.js · Next.js · REST APIs · Machine Learning
1
9
Built a production-ready deep learning system that classifies 38 plant disease states across 14 species with 99.98% accuracy.
The system uses EfficientNetV2S as its backbone with CBAM attention mechanisms that force the model to focus strictly on disease markers like necrotic spots and chlorosis patterns — ignoring background noise completely.
Trained on 87,000+ images from the PlantVillage dataset with Test-Time Augmentation that averages three augmented views per inference, boosting prediction confidence with minimal latency impact.
Integrated Gemini AI as a pre-validation gate ensuring the model only processes actual leaf images, virtually eliminating false positives.
Tech Stack: TensorFlow · EfficientNetV2S · CBAM · Gemini AI · Python · Computer Vision
1
12
Built a fully functional 24/7 AI personal assistant using N8N and GPT-4o with zero traditional coding.
The system handles real-time weather, news, and stock tracking, manages full Google Workspace including emails, calendar, and Drive, remembers user preferences using Google Sheets memory, delivers automated morning briefings at 8 AM daily, and monitors website uptime with instant alerts.
Tech Stack: N8N · GPT-4o · Claude · Google Workspace · OpenWeather · SerpAPI · Alpha Vantage · CoinGecko
0
16
AI-Powered Hiring Automation System
Built a complete end-to-end recruitment automation using N8N + OpenAI + Google Workspace.
The Problem:
Companies waste weeks manually screening CVs, scoring candidates inconsistently, and sending individual emails one by one.
What I Built:
Two-workflow system that runs 24/7 automatically:
Workflow 1 — Screening Pipeline
Captures applications from Google Forms in real-time
Downloads and reads resumes from Google Drive
AI evaluates each candidate with role-specific criteria
Scores 0-100 and classifies Strong/Average/Weak
Stores everything in a Master Google Sheet
Workflow 2 — Email Pipeline
Runs every 3 minutes automatically
Strong candidates → Interview invite + Calendar booking
Average candidates → Manager notification
Weak candidates → Professional rejection email
Result: Hiring cycle from weeks → hours. Zero manual work.
Stack: N8N · OpenAI · Google Forms · Gmail · Google Calendar · JavaScript